Softening and Melting Characteristics of Self-fluxed Pellets with and without the Addition of BOF-slag to the Pellet Bed

Transcription

1 , pp Softening and Melting Characteristics of Self-fluxed Pellets with and without the Addition of BOF-slag to the Pellet Bed Guangqing ZUO Division of Process Metallurgy, Luleå University of Technology, Luleå, Sweden. (Received on June 5, 2000; accepted in final form on August 18, 2000) When using 100% self-fluxed pellets in the blast furnace burden, top charged fluxes, especially the BOFslag, may cause irregularities in operation. The major reason has been theoretically attributed to the problematic slag formation in the furnace. As the melting of pellets is the first step of the slag formation process, the softening and melting properties of LKAB s self-fluxed pellets with and without addition of fluxes are studied experimentally. The results show that the softening and melting properties of the two types of LKAB s self-fluxed pellets are quite suitable for blast furnace operation. Contrarily, the melting-down characteristics of BOF-slag are variable and harmful to the slag formation under a reducing atmosphere. Adding 5% BOF-slag to the selffluxed pellets can considerably worsen the softening and melting properties of pellets. It can either increase the softening and melting temperature interval substantially or cause the precipitation of solid phases, mainly di-calcium silicates, in the slag. It is quite likely that the resulting slag will therefore become very viscous, even entirely blocking the melting down of the pellets up to a temperature C. KEY WORDS: blast furnace; self-fluxed pellets; slag formation. 1. Introduction The use of olivine pellets, developed jointly by LKAB and SSAB, Sweden, has considerably improved blast furnaces performance in the last 20 years at SSAB Luleå, Sweden. 1) To obtain even better performance in the blast furnace, new type of self-fluxed pellets, with less olivine and an addition of limestone, are under development. Besides the higher iron content, the self-fluxed pellets, according to the laboratory tests, have higher reducibility and better softening and melting characteristics. 2) But the industrial trials using the new self-fluxed pellets and topcharged fluxes, conducted at SSAB Luleå, have not shown further enhancement of the production. On the contrary, it did cause some irregularities in the furnace operation, 3) e.g. high pressure drop cross the burden column, unstable permeability and difficulties in tapping etc. One of the major reasons has been theoretically attributed to the worsening of the softening and melting properties of the pellet in the furnace, resulting from the top-charging of the fluxes together with pellets. 4) Conventionally, the major slag formers used at SSAB Luleå are BOF-slag and limestone, top-charged together with pellets. The amount of the BOF-slag added is about 3% of the amount of pellet. Theoretical analyses conducted by Professor Ma 4) have shown that top-charged BOF-slag can improve slag formation when using olivine pellets, but does cause problems when using 100% self-fluxed pellets. For the latter, the top-charged BOF-slag with the primary slag phases from the pellets can form high-basicity bosh slag, especially when segregation of BOF-slag occurs. The basicity CaO/SiO 2 B 2 of bosh slag can reach about 1.8 with a high melting point when 6% BOF-slag is added. The excessive slag basicity can cause an increase in melting temperature of the burden, resulting in a thicker cohesive zone and the consequent high resistance of the burden to gas flow. Re-solidification of the slag may also occur. As a result, irregularities as mentioned above, and even hanging and slip can occur. The smooth operation of the furnace can be interrupted. Actually, the softening and melting of slag in the blast furnace is a quite complex process. It involves the iron oxides reduction, liquid formation, deformation, carbonization and melting processes and various chemical reactions. Besides the composition and quantity of gangue and the structures of pellets, the process also depends on the quantity of wustite and the quantity of the additional fluxes. In other words, softening and melting characteristics of the burden are strongly influenced by the extent of reduction of pellets and the chemical compositions of bosh slag. To improve the quality of the blast furnace burden considerable effort has been invested in studies of softening and melting properties. A number of laboratory tests have been developed to obtain information about softening and melting characteristics of iron-bearing materials. 5 11) The results of these tests are used to estimate potential effects on the characteristics of the cohesive zone in the furnace. In general, these tests were performed using iron-bearing materials only without the addition of slag formers to test samples. Therefore, these test results may provide rather good indications of the softening and melting characteristics of iron-bearing materials, but cannot appropriately show the ISIJ

2 Fig. 1. Schematic view of experimental apparatus. melting properties of the burden in the furnace, mixed with some slag formers, especially when top-charging fluxes together with the self-fluxed iron-bearing materials. In order to reveal the effect of fluxes on the melting characteristics of the self-fluxed pellets, being developed at LKAB, softening and melting tests of these two types of self-fluxed pellets with and without the fluxes have been carried out in the laboratory. This paper reports the experimental results. The causes of irregularities occurring in conjunction with 100% use of self-fluxed pellets will also be discussed. 2. Experimental Technique The experimental apparatus is shown in Fig. 1. The furnace is heated electrically by U-shape-Super Kanthal with a heating zone of about 800 mm in height. The maximum working temperature can reach C. The crucible (40 mm i.d. 85 mm), used, made of graphite or alumina as shown in Fig. 2, has 6 holes of 6 mm in diameter on the bottom, allowing the dripping of the molten materials and inlet of gas through the sample bed. A balance mounted with a container is installed beneath the reaction tube for collecting and weighing the dripped materials. Throughout every test, a constant load of 0.9 kg/cm 2 is applied to sample bed by means of a pneumatic cylinder. The pressure given by the cylinder and the nitrogen gas flow through sample bed are regulated via a computer PC 1. The bed contraction, the pressure drop across the sample bed, the mass dripped off as well as the temperature in the course of an experiment are measured with corresponding devices and recorded by a second computer PC 2. After reaching the maximum test temperature, the sample is cooled down to ambient temperature. The whole test procedure is conducted under the protection of nitrogen gas with a flow rate of 7 l/min. Two types of self-fluxed pellets, developed by LKAB, were tested. Pellet A has a slightly higher CaO content, but Fig. 2. Placement of samples in the furnace tube. a lower MgO content than pellet B: The silica and alumina contents are similar. The chemical analysis of the raw pellets and slag formers used are shown in Table 1. Before the softening and melting tests, pellets were prereduced to reduction degrees in percentage 60, 75 and 90 respectively under conditions of 850 C and 55%N 2 40%CO 5%H 2 reducing gas (Cracks of pellets during the reduction could then be avoided to a large extent). BOFslag used in these experiments was also pre-reduced under the same reduction conditions for 90 min. However, the reduction degree was only about 15%. BOF-slag, coke breeze, burnt dolomite and limestone were taken from the SSAB Luleå Works, Sweden. For testing the impact of BOF-slag on the softening and melting properties of pellets, three different placements of BOFslag in the pellet bed were used, as shown in Fig. 3. Considering the ratio of the BOF-slag at No. 2 BF, SSAB, Luleå, additions of BOF-slag to pellet samples were set at 5 and 10% of the amount of pellets used, corresponding to the uniform distribution and segregation of BOF-slag in the furnace. The experimental conditions used in the study are listed in Table 2. For characterizing the softening and melting behaviors of the samples, following two indices have been derived ISIJ 1196

3 Table 1. Chemical analysis of the test samples. Fig. 3. Three different placements of BOF-slag in pellets bed. Table 2. Summary of test conditions. T 50 : softening temperature, the temperature at which the bed contraction reaches 50% of the original sample bed height; T sd : start-of-dripping temperature, the temperature at which the first droplet of the melting material drops to the collector, which also serves as the melting temperature in the study. Figure 4 shows examples of the softening and melting temperature of pellet A. 3. Experimental Results 3.1. Softening and Melting Properties of Pre-reduced Pellets and BOF-slag The softening and melting test results of pre-reduced pellet A of 75% pre-reduction degree and BOF-slag using graphite and alumina crucibles are presented in Table 3. The results showed that the pellet and the BOF-slag behaved differently when using different crucibles, depending on the presentation of carbon to the samples. When a graphite crucible was used or the pellets were sandwiched using two layers of coke, the softening and melting temperature of pellets were the same, about C and C respectively. But the BOF-slag did not melt down at all until reaching a temperature C. When an alumina crucible was used without presentation of carbon, pellet samples did not melt down when a temperature of C was reached. Pellet samples deformed, but still had iron shells (Photo 1). The BOF-slag started dripping at a temperature of about C, which is in accordance with its Fig. 4. Definition of the softening and melting temperature. melting point. 12) Figures 4 and 5 show the deformation process of pellet A and BOF-slag through the tests. These tests indicate that the coke or carbon has an important role in the softening and melting of the pellets and BOF-slag. To obtain softening and melting properties relevant to the conditions in the furnace, the samples should be therefore sandwiched using two layer of coke for the latter tests, which is relevant to the conditions in the furnace Influence of the Placement of BOF-slag Softening and melting tests of pre-reduced pellets A with 75% reduction degree and BOF-slag distributed in three different ways in the pellet bed were carried out. The maximum temperature reached in these experiments was C. Some of the test results are shown in Fig. 6. These results indicated that the influence of the BOF-slag on the softening and melting properties of pellets was largely dependent on its distribution in the pellet bed. When it was evenly distributed in the pellet bed, the melting temperature was considerably increased. Even when a temperature of C was reached, dripping did not occurr. The pellets and the BOF-slag formed one mixed segment of metallic irons and slag as shown in Photo 2. The melting temperature of the samples was higher than C. When BOF-slag in the pellet bed was segregated, the melting temperature was increased too, but to a lesser extent compared to that above. However, some white-gray powder was found in the sample bed after the experiment ISIJ

4 Table 3. Softening and melting test results of pellets with 75% pre-reduction degree and BOF-slag. Photo 1. Samples obtained after the softening and melting test. Photo 2. Bulk of metallic iron and slag. Fig. 5. Deformation of BOF-slag. Fig. 7. X-ray diffraction analysis of the powder collected from sample bed. Fig. 6. Changes of the softening and melting of pellets with the placement of BOF-slag (5%). When a BOF-slag layer was placed beneath the pellet layer, the softening and melting temperature of the pellets was not affected. However, the white-gray powder was also found in the sample bed after the experiment. X-ray diffraction analyses showed that the white-gray powder contained di-calcium silicate (Ca 2 SiO 4, C), and other complex compounds, e.g. 3CaO Al 2 O 3, 3CaO SiO 2 etc., as shown in Fig. 7. TGA test showed that the melting temperature of the powder was higher than C. These results might suggest that the compounds of high melting points, e.g. di-calcium silicates may constitute the major composition of the powder Influence of the BOF-slag and the Reduction Degree of Pellet Addition of BOF-slag to the pellet bed could considerably change the melting properties of pellets. 5% or 10% addition of BOF-slag evenly to the pellet bed (75% prereduction degree) could increase the melting temperature from about C to a level above the maximum experimental temperature of C. The softening temperature increased by about 25 C only, as shown in Fig. 8a). As a result, the softening and melting temperature interval could 2000 ISIJ 1198

5 Fig. 8. Softening and melting temperature of test samples. Table 4. Softening and melting properties of pellet B. Fig. 9. Variations in pressure drop with increase in temperature. increase. When BOF-slag was placed unevenly in the sample bed or as a layer beneath the sample bed, it increased the melting temperature by about 50 C, but generated solid particle with very high melting points as mentioned above. Evenly, the solid particle in slag can worsen its fluidity. BOF-slag also increased the pressure drop cross the sample bed. Figure 9 shows variations of the pressure drops cross the sample bed with the increase in temperature. With and even addition of 5% BOF-slag to pellet bed, the maximum pressure drop increased by about 10%. But the temperature, at which the pressure drop reaches maximum, increased considerably by about 80 C, indicating higher startof-dripping temperature of the samples. The effect of the pre-reduction degree of pellets used on the softening temperature of the samples was quite evident, irrespective of the addition of BOF-slag to the sample bed. For instance, increasing the pre-reduction degree from 60 to 90% could result in an increase in the softening temperature of about 70 C, Fig. 8a). The influence of pre-reduction degree on melting temperature could vary, depending on the presentation of BOFslag. When there was no addition of BOF-slag, the melting temperature increased about 20 C with the increase of the pre-reduction degree of pellet from 60 to 90%. When adding 5% or more BOF-slag evenly to the sample bed, the melting temperature of the sample was higher than the maximum experimental temperature used, regardless of the pre-reduction degree of the pellets Effect of Some Other Additions Tests of adding limestone and burnt dolomite to pellet bed were conducted as well. The basicity B 2 CaO/SiO 2 or B 3 (CaO MgO)/SiO 2 of the slag was adjusted to the same level as that of an addition of 5% BOF-slag. Pellet A with a reduction degree of 75% was used and slag formers were evenly distributed in the pellet bed. The results showed that limestone or burnt dolomite could increase softening and melting temperature by about 30 C and 25 C and respectively. The influences of these two types of slag formers were less than that of BOF-slag. The differences might be partly due to the reduction of the BOF-slag during heat-up. Gray-white powders could also be found in the sample bed after the experiment. However, the amount was not enough to warrant further examination Softening and Melting Properties of Pellet B Softening and melting tests using pellet B with a pre-reduction degree of about 75% were carried out. The test results are shown in Table 4. These results showed that the phenomena occurring in these tests were quite similar to what was observed in tests using pellet-a, except for the higher softening temperature of pellet B. Evenly distributed BOF-slag in the pellet bed could entirely block the meltdown of the pellet, while segregation of BOF-slag in the pellet bed increased the softening and melting temperature interval, caused more residuals and generates solid phases in the sample bed of high melting temperature. 4. Discussion 4.1. Melting Properties of Self-fluxed Pellets and BOFslag The test results reveal that the two types of self-fluxes pellets have quite good softening and melting properties. The relatively high softening temperature C, lower melting temperature C and the narrow softening and melting temperature interval can facilitate to the formation of a low and thin cohesive zone in the furnace, enlarging the thermal reserve zone and reducing the resistance to gas flow. Uses of these two types of self-fluxed pellets should give potential for the further improvement in blast furnace operation. The melting properties of BOF-slag were variable, strongly depending on the presentation of carbon. These ISIJ

6 phenomena may be due to the reduction of the FeO in the slag by carbon. Chemical analysis of the BOF-slag after the tests up to C showed that the FeO and Fe 2 O 3 in the slag were only 5.3% and 1.5%, respectively, while the originals were 16.4% and 8.08%, respectively. Melting point tests showed that the melting temperature of the BOF-slag after the tests was higher than C, which was much higher than that about C, of the original BOF-slag. Therefore, it can be concluded that when the top-charged BOF-slag reaches the cohesive zone, it will not melt down before forming bosh slag with the primary slag from the core of the pellets Causes of Problematic Melting Process When Adding Fluxes to Pellet Bed High basicity and low FeO content of the slag formed during the melting process of pellets and BOF-slag can be the major reason for the problematic melting process. In the course of heating-up to the maximum experimental temperature, the primary slag from pellets with the BOF-slag forms new slag, which corresponded to the bosh slag at the cohesive zone in the blast furnace. Table 5 shows the estimated chemical compositions of the primary and BOF slag, as well as bosh slag with assumed different FeO contents in the slag. Bosh 1 3 corresponds to 5% addition of BOF-slag to pellets and bosh 4 to 10% addition. It is assumed that all of gangues in the pellets and all of BOF-slag enter into the bosh slag. The primary slag had a basicity B2 of about 1.0. Its melting point is located in a low melting point region as shown Table 5. Chemical composition of slag. in Fig. 10a)-A. Considering the presentation of FeO in the primary slag, its melting point would become even lower. When the basicity of slag is about 1.0, changes in the FeO content in the slag will not considerably affect the melting temperature, Fig. 10b). Obviously, the primary slag does not impair the melting of the pellets. However, due to the high CaO content of BOF-slag, the new slag formed during the melting process had a basicity B 2 about 1.87 and B 3 about 2.43, as shown in Table 5. It can be seen from the diagram of Fig. 10a)-A1 and Fig. 10b)-C that this slag has a much higher melting point than the primary slag. Increasing the basicity of the slag or decreasing the FeO content will further increase its melting point. The viscosity at the same temperature levels will also increase when the FeO content is decreased. The fluidity will deteriorate, causing difficulties in the separation of the metallic iron and slag, even entirely blocking dripping. In the cases of the segregation of the BOF-slag or adding more BOF-slag to the sample bed, the basicity of the slag became even higher in some local region of the sample bed. Assuming double BOF-slag locally in the experiment or adding 10% BOF-slag makes B 2 and B 3 of the newly formed slag about 2.36 and 3.07, respectively. The melting point of the slag increases even further as shown in Fig. 10a)-A2. Considering the reduction of FeO in the slag, the melting property of the newly formed slag will be considerably worsened. Di-calcium silicate, which has a melting point of C will be produced in the new slag. Due to the low FeO content, the precipitation of di-calcium silicates in the slag can occur. The generation of the white powders collected from the sample beds after the tests has demonstrated this phenomenon. Di-calcium silicates were also found in the blast furnace slag when industrial tests using pellet A 3) were performed. The solid phases in the slag generated in the course of the melting have an even more detrimental effect on the melting properties of pellets, hindering melting of pellets and separation of metallic iron from slag phase. Fig. 10. Phase diagrams ISIJ 1200

7 4.3. Influence of the BOF-slag on Softening and Melting Temperature The effect of the BOF-slag on softening temperature of the sample bed was probably due to the higher softening temperature of the BOF-slag. The different effect of three different placement of BOF-slag on the melting temperature of the sample might result from the different slag formation process. Test results have demonstrated that the softening temperature of the BOF-slag was higher than C when using graphite crucible or conjecturing with coke. The softening temperature of the sample was defined as the temperature at which, the contraction ratio of the sample bed reached 50% of its original height. Addition of BOF-slag to the pellet bed could decrease the contraction rate of the sample bed, resulting in the increase of the softening temperature. Due to the difference in the particle size between the pellets and BOF-slag, the BOF-slag could distribute in the voidage of the pellet bed. Apparently, a 5 10% BOF-slag does not change the structure of the pellet bed, nor the softening temperature of the sample. During the melting process, primary slag from the pellet cores and BOF-slag form new slag. When BOF-slag was evenly distributed in the sample bed, the newly formed slag had a high basicity and consequently low fluidity and even distribution in the sample bed. Because of its low fluidity it could entirely block the separation of metallic iron from slag and the occurrence of dripping until reaching the maximum experimental temperature of C. When BOFslag was unevenly distributed, metallic iron could separate from the slag in some local region of the sample. Although the melting temperature of the sample increased, dripping could occur. When the BOF-slag layer was placed beneath the pellet bed, metallic iron could separate from the slag prior to the formation of new slag. Therefore, it appeared that the melting temperature was not affected at all. Actually, for the later two cases, the newly formed slag could have even higher basicity and poorer fluidity in some local region due to the uneven slag formation process, resulting in the easy precipitation of solid particles in the slag. The powder generated during the tests has proved that the uneven distributed BOF-slag in the sample bed could produce compounds of high melting points in the slag Melting Process of Pre-reduced Pellets Three melting-down mechanisms of pellets have been proposed. 10) It is generally accepted that the core of pre-reduced pellets will melt and exude out from the center of pellets prior to the melting-down of the iron shell. In this study, it seemed that the flowing-out of the primary slag formed in the core happened mainly when the reduced iron shell melted, or after it melted. The exudation prior to the melting-down also occurred, but to a much lesser degree. Tests showed that pre-reduced pellets of 75% pre-reduction degree without contacting carbon would not melt down when a temperature of C was reached. Although deformation occurred, pellets still had the iron shell, Photo 1. Another interrupted tests (interrupted at C) showed that when carbon (from coke or graphite crucible) was presented to pellets, the sides of the iron shells of pellets, adjoining the coke or graphite, would melted down, Photo 3. Photo 3. Pellets after interruption of the softening and melting test. But the opposite sides still had the round shape with a shaped hollow left by the flowing-out of the slag formed in the core. The pellets that did not come in contact with carbon kept their round shape. These phenomena might indicate that flow-out of primary slag from the core of pellets prior to the melting of the iron shell did not occur to any significant degree. The melting of the iron shell and the flowing-out of the primary slag gradually proceeded layer by layer throughout the pellet bed. The liquid slag was further reduced, forming the metallic iron, gathering in the crucible and finally dripping out The Influence of Top-charged Fluxes on Blast Furnace Operation When Using the Self-fluxed Pellet The top-charged BOF-slag may cause irregularities in blast furnace operation when using the new type of selffluxed pellets, for instance, the low fluidity of slag, the problematic tapping process, the increase of the thickness of the cohesive zone and high resistance to gas flow, etc. The laboratory tests have demonstrated the difficulties of the separation of the iron from slag, and the generation of the solid particles when adding only 5% BOF-slag or other fluxes to the pellet bed. This type of slag may appear in the bosh region in the blast furnace. The liquid primary slag from pellets and the BOF-slag in the bosh region can form new slag before absorbing the acid elements of coke or coal. Calcium silicates and some other compounds of very high melting temperature can be generated due to the high basicity of the newly formed slag. The bosh slag will become very viscous, and even re-solidify. The solid particles will be present in the slag even when the slag absorbs the acid element of coke or coal in the lower part of the furnace. As a result, the cohesive zone may become thicker, giving the higher resistance to gas flow. The final slag can be viscous, hindering the separation of the metallic iron from slag and the chemical reactions between the liquid iron and slag in the hearth. The smooth running of the furnace will be disrupted. The quality of the iron will deteriorate and the tapping process will be problematic. These phenomena have been demonstrated by the industrial trial, ISIJ

8 conducted at SSAB Luleå Works, Sweden using 100% selffluxed pellets A of type A. 3) 5. Conclusions The softening and melting characteristics of the selffluxed pellets has been tested with and without the addition of BOF-slag to the sample bed. The results can be summarized as follows: ( 1 ) The softening and melting properties of the two types of self-fluxed pellets developed at LKAB, Sweden, are quite good and suitable for blast furnace operation. The melting characteristics of BOF-slag are variable, strongly depending on the presence of carbon. BOF-slag may not melt down before forming new slag with the primary slag of the pellets in the blast furnace. ( 2 ) Addition of BOF-slag to a bed of self-fluxed pellets can cause a problematic slag formation process. The main reasons are the formation of high basicity slag by BOF-slag and primary slag from the pellet cores and the reduction of FeO. Consequently, top-charging of BOF-slag when using self-fluxed pellets may result in serious irregularities in blast furnace operation. ( 3 ) The carbonization of the iron shell of pellets plays an important role in the melting of the pellets. The melting of the reduced iron shell and the flowing out of the primary slag generated in the core may occur simultaneously. The exudation of the primary slag from the pellet cores prior to the melting of the iron shell may also occur but to a much lesser extent in this study. Acknowledgements The author wishes to thank Prof. Jitang Ma retired from Luleå University of Technology and Prof. Bo Björkman of Luleå University of Technology for their guidance, help and encouragement. The financial support of the Swedish Steel Producers Association (Jernkontoret) is gratefully acknowledged. Thanks are due to SSAB Tunnplåt, Sweden, LKAB Sweden for their strong support to this project and all the experimental materials provided. Mrs Birgitta Nyberg and Mr. Ola Kask and other colleagues at the Division of the Process Metallurgy, Luleå University of Technology are also gratefully acknowledged for their assistance in the research work. I also wish to express my thanks to the committee of the project Injection into Blast Furnace. References 1) M. Brämming, M. Hallin and G. Zuo: Proc. of the 54th Ironmaking Conf., ISS, Warrendale, PA, (1995). 2) N. Sandberg, O. Löfgren and M. Tottie: Proc. of the 55th Ironmaking Conf., Vol. 55, ISS, Warrendale, PA, (1996), ) L Hooey and B Sundelin: 1998 ICSTI/Ironmaking Conf. Proc., Vol. 57, ISS, Warrendale, PA, (1998), ) J. Ma: ISIJ Int., 39 (1999), No. 7, ) Y. Ishikawa M. Kase, M. Sasaki, K. Satoh and S. Sasaki: Proc. AIME Ironmaking Conf., Vol. 41, ISS-AIME, Warrendale, PA, (1982), 80. 6) Y. Yamaoka, H. Hotta and S. Kajikawa: Trans. Iron Steel Inst. Jpn., 22 (1982), ) G. Clixby: Ironmaking Steelmaking, 7 (1980), 68. 8) L. Bentell and L. Norman: Proc. AIME Ironmaking Conf., Vol. 41, ISS-AIME, Warrendale, PA, (1982), ) L.-H. Hsieh and K.-C. Liu: 2nd Int. Cong. on the Science and Technology of Ironmaking and 57th Ironmaking Conf. Proc., ISS, Warrendale, PA, (1998), ) G. Clixby: Ironmaking Steelmaking, 13 (1986), ) V. J. Ritz and H. A. Kortmann: 1998 ICSTI/Ironmaking Conf. Proc., ISS, Warrendale, PA, (1998), ) L. Sundqvist Ökvist: Melting Properties of BOF-Slag and Their Influence on the Slag Formation, Mater s Thesis, Division of Process Metallurgy, Luleå University of Technology, Sweden, (1998), ) E. M. Levin, C. R. Robbins and H. F. McMurdie: Phase Diagrams for Ceramists, The American Ceramic Society, INC., Columbus, OH (1969). 14) E. M. Levin, C. R. Robbins and H. F. McMurdie: Phase Diagrams for Ceramists, Vol. I, The American Ceramic Society, INC., Columbus, OH (1964) ISIJ 1202